Embedding to Reference t-SNE Space Addresses Batch Effects in Single-Cell Classification

Abstract

Dimensionality reduction techniques, such as t-SNE, can construct informative visualizations of high-dimensional data. When working with multiple data sets, a straightforward application of these methods often fails; instead of revealing underlying classes, the resulting visualizations expose data set-specific clusters. To circumvent these batch effects, we propose an embedding procedure that takes a t-SNE visualization constructed on a reference data set and uses it as a scaffold for embedding new data. The new, secondary data is embedded one datapoint at the time. This prevents any interactions between instances in the secondary data and implicitly mitigates batch effects. We demonstrate the utility of this approach with an analysis of six recently published single-cell gene expression data sets containing up to tens of thousands of cells and thousands of genes. In these data sets, the batch effects are particularly strong as the data comes from different institutions and was obtained using different experimental protocols. The visualizations constructed by our proposed approach are cleared of batch effects, and the cells from secondary data sets correctly co-cluster with cells from the primary data sharing the same cell type.

Predicting Endocrine Disruption Using Conformal Prediction – A Prioritization Strategy to Identify Hazardous Chemicals with Confidence

“…t-SNE is an unsupervised nonlinear probabilistic exploratory and visualization algorithm, which embeds high-dimensional data for visualization in a low-dimensional space. 36 In brief, the principle behind t-SNE is calculation and comparison of probabilities of proximity in higher- and lower-dimensional space. This comparison is the premise for visualization, where the differences are attempted to be minimized in a lower-dimensional space, using local minima by applying a gradient descent.…”

“…t-SNE is an unsupervised nonlinear probabilistic exploratory and visualization algorithm, which embeds high-dimensional data for visualization in a low-dimensional space . In brief, the principle behind t-SNE is calculation and comparison of probabilities of proximity in higher- and lower-dimensional space.…”

“…A detailed explanation of the method is presented by van det Maaten and Hinton. 37 For the purposes of this work, t-SNE was performed, following the protocol openTSNE (version 0.3.11) 36 in Python environment (version 3.7) using conformal p-values as input features for the studied data set. Tanimoto coefficients are scores that represent similarity between sets of elements such as fingerprints in binary.…”

Sub-Cluster Identification through Semi-Supervised Optimization of Rare-cell Silhouettes (SCISSORS) in Single-Cell Sequencing

“…Gene expression is normalized using regularized negative binomial regression as implemented in SCTransform (18), which also provides variance estimates for each gene. The top 4,000 highly variable genes (HVGs) are used to reduce the dimensionality of the gene-cell matrix through PCA, and the first 20 principal components are used as an initialization for a two-dimensional Fast Fourier Transformaccelerated t-SNE embedding (19,20). The cells are clustered in PCA space using Louvain modularity optimization after being embedded in a shared nearest-neighbors graph.…”

Embedding to Reference t-SNE Space Addresses Batch Effects in Single-Cell Classification

“…We extend the t-SNE algorithm to enable the embedding of new data points to a reference embedding [37]. We adapt the standard t-SNE formulation from Eqs.…”